430413 - Port Mozilla to GDK

Status: RESOLVED WONTFIX

(Core :: Widget: Gtk, defect)

Description

[Jiny Sung]

User-Agent:       Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; InfoPath.2; .NET CLR 1.1.4322)
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b5) Gecko/2008032619 Firefox/3.0b5

If you look into Firefox3's APIs, you can find out that some of them directly use GDK/X11 methods. It generates dependency with X11.
Actually Firefox3 supports only X11 backend on Linux platform.
However, there is a possibility that several different back-end are supported. 
This bug intends to port Mozilla regardless of back-end. Therefore, we need to remove specific X dependency and raise it to GDK which acts as a wrapper around the low-level drawing and windowing functions provided by the underlying graphics system. By converting X APIs to GDK's, it will be more flexible. A wiki page (http://wiki.mozilla.org/Mobile/DFBPorting) and bug 422221 explaining DirectFB porting strategy have been opened by my coworker, Mohan Reddy Duggi. Compared with previous bug, this bug makes Firefox act under GDK (DirectFB as well as X11) without defects.

Reproducible: Always

Comment 1

[Jiny Sung]

Attachment: gfx.patch for GDK porting

Three files (nsSystemFontsGTK2.cpp, nsThebesDeviceContext.cpp and gfxPlatformGtk.h) in gfx (except surface) module are modified in order to remove direct X dependency.

Comment 2

[Jiny Sung]

Attachment: toolkit.patch for GDK porting

This patch will be cleaned up X dependency of the two files (nsGTKRemoteService.h and nsGTKRemote.cpp) in toolkit module.

Comment 3

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Comment on attachment 317215 [details] [diff] [review]
gfx.patch for GDK porting

This looks fine, but please don't nest the #ifdefs.. e.g.,

#ifdef MOZ_ENABLE_GTK2
#ifdef MOZ_X11

should just be

#if defined(MOZ_ENABLE_GTK2) && defined(MOZ_X11)

I'll take a look at the other patch in a bit.

Comment 4

[Benjamin Smedberg]

Comment on attachment 317234 [details] [diff] [review]
toolkit.patch for GDK porting

Caillon, could you look at this? I don't know much about gdk, and I want to make sure that switching from X to GDK will retain backwards compatibility with older clients.

Comment 5

[Christopher Aillon (sabbatical, not receiving bugmail)]

Comment on attachment 317234 [details] [diff] [review]
toolkit.patch for GDK porting

>-  XInternAtoms(GDK_DISPLAY(), XAtomNames, NS_ARRAY_LENGTH(XAtomNames),
>-               False, XAtoms);
>+  for(int i=0 ; i<NS_ARRAY_LENGTH(GdkAtomNames) ; i++)
>+  {
>+  GdkAtoms[i] = gdk_atom_intern_static_string(GdkAtomNames[i]);
>+  }

So, there is no GDK equivalent to XInternAtoms.  This change specifically will regress us in terms of roundtrips to the X server, not to mention the additional calls to gdk_intern_atom() below.  On the local machine, this is probably not too bad.  But a big use case of the remote service is from remote machines, and this will significantly hurt performance where roundtrips are not cheap...

Comment 6

[Jiny Sung]

As pointed out above, it's better to leave original source code in case of using X backend. So, I tried to divide this part into two branches (when using X11 or not). 
However, I wonder whether it demands our serious consideration or not. Because Firefox's remote control feature seems to be only effective through X properties.  
This feature is included in 'toolkit.patch for GDK porting'. Without using X backend, I guess that 'remote' directory which is a higher rank of 'nsGTKRemoteService.cpp' and 'nsGTKRemoteService.h' is excluded.     

Comment 7

[Jiny Sung]

Comment on attachment 317215 [details] [diff] [review]
gfx.patch for GDK porting

In order to make the patch be coincident with Bugzilla's rules

Comment 8

[Jiny Sung]

Attachment: The patch was modified with Bugzilla's style.

The previous patch was modified with Bugzilla's style.
You can simply comfirm the details of the changed part.

***Old version***
#ifdef MOZ_ENABLE_GTK2
#ifdef MOZ_X11

***New version***
#if defined(MOZ_ENABLE_GTK2) && defined(MOZ_X11)

Comment 9

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Comment on attachment 320147 [details] [diff] [review]
The patch was modified with Bugzilla's style.

This looks fine, thanks!

Comment 10

[Christopher Aillon (sabbatical, not receiving bugmail)]

Comment on attachment 317234 [details] [diff] [review]
toolkit.patch for GDK porting

minusing based on my previous comment 5

Comment 11

[Jiny Sung]

Comment on attachment 317234 [details] [diff] [review]
toolkit.patch for GDK porting

Consider 'XRemote' service

Comment 12

[Andrew Overholt [:overholt]]

I'm guessing at a better component here.

Comment 13

[BugBot [:suhaib / :marco/ :calixte]]

The bug assignee didn't login in Bugzilla in the last 7 months.
:stransky, could you have a look please?
For more information, please visit auto_nag documentation.